Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
2978 V. Vaithyanathan et al. / Acta Materialia 52 (2004) 2973–2987<br />
energetics is applicable only for substitutional fcc-based<br />
configurations. It can be used for calculat<strong>in</strong>g the free<br />
energy <strong>of</strong> an fcc-based compound (e.g., a structure that<br />
can be represented by a substitutional decoration <strong>of</strong> A<br />
and B atoms on the sites <strong>of</strong> an fcc lattice, such as an<br />
ordered L1 2 or L1 0 ), but cannot be used for calculat<strong>in</strong>g<br />
the free energy <strong>of</strong> h 0 , which has a (distorted) CaF 2<br />
crystal structure. Hence, the h 0 energy is obta<strong>in</strong>ed from<br />
direct first-pr<strong>in</strong>ciples calculations at T ¼ 0 K, coupled<br />
with the vibrational entropy <strong>of</strong> h 0 , which has been found<br />
to be unexpectedly important for this phase [1]. Additionally,<br />
by perform<strong>in</strong>g po<strong>in</strong>t defect calculations <strong>of</strong> supercells<br />
<strong>of</strong> h 0 and <strong>in</strong>sert<strong>in</strong>g the energetics <strong>in</strong>to a lowtemperature<br />
expansion, we f<strong>in</strong>d that the configurational<br />
entropy <strong>of</strong> h 0 is small, and we do not consider h 0 <strong>of</strong>fstoichiometry.<br />
The h 0 formation enthalpy was calculated<br />
from first-pr<strong>in</strong>ciples as )195.8 meV/atom at <strong>Al</strong> 2 <strong>Cu</strong><br />
stoichiometry (X <strong>Cu</strong> ¼ 1=3). The computed (formation)<br />
vibrational entropy is )0.62k B [1]. Hence, the free energy<br />
<strong>of</strong> h 0 as a function <strong>of</strong> temperature is given by<br />
DF h<br />
0ðT Þ¼ 195:8 þ 0:62k B T ð<strong>in</strong> meV=atomÞ: ð15Þ<br />
We note that without the <strong>in</strong>clusion <strong>of</strong> the h 0 vibrational<br />
entropy term, the calculated equilibrium solubility <strong>of</strong> <strong>Cu</strong><br />
<strong>in</strong> solid solution is extremely small. For easier visualization,<br />
the free energy is scaled by a l<strong>in</strong>ear term <strong>in</strong><br />
composition ½ 3X <strong>Cu</strong> DF h<br />
0ðT ÞŠ, such that the rescaled h 0<br />
free energy at X <strong>Cu</strong> ¼ 1=3 is zero (Fig. 4(c)). This l<strong>in</strong>ear<br />
scal<strong>in</strong>g <strong>of</strong> the free energy does not affect the determ<strong>in</strong>ation<br />
<strong>of</strong> equilibrium composition through the common<br />
tangent construction.<br />
3.2. Interfacial energy<br />
In a system where the precipitate and matrix phases<br />
are substitutional rearrangements <strong>of</strong> atoms on the same<br />
lattice type, and form perfectly coherent <strong>in</strong>terfaces (e.g.,<br />
GP zone phases), one could calculate the <strong>in</strong>terfacial free<br />
energy from a Monte Carlo simulation coupled with<br />
thermodynamic <strong>in</strong>tegration, analogous to the procedure<br />
described for bulk free energies. <strong>Al</strong>ternatively, T ¼ 0K<br />
values may be obta<strong>in</strong>ed from direct first-pr<strong>in</strong>ciples supercell<br />
calculations (without us<strong>in</strong>g a CE). h 0 precipitates<br />
are partially coherent with different crystal structures for<br />
precipitate and matrix. Hence, it is not amenable to the<br />
CE method. Therefore, we extract the T ¼ 0 K <strong>in</strong>terfacial<br />
energies directly from first-pr<strong>in</strong>ciples supercell calculations.<br />
We next describe how the <strong>in</strong>terfacial energies<br />
are extracted from supercell energetics, and separated<br />
from the coherency stra<strong>in</strong> energetics.<br />
We beg<strong>in</strong> by consider<strong>in</strong>g an N-atom coherent supercell<br />
conta<strong>in</strong><strong>in</strong>g an <strong>in</strong>terface between two materials, A<br />
and B (<strong>in</strong> the case <strong>of</strong> this work, A ¼ <strong>Al</strong> and B ¼ <strong>Al</strong> 2 <strong>Cu</strong><br />
h 0 ). For simplicity, we consider the case <strong>of</strong> cells compris<strong>in</strong>g<br />
equal amount <strong>of</strong> A and B. The energy <strong>of</strong> a such<br />
an A N=2 B N=2 supercell can be separated <strong>in</strong>to two components<br />
[27]: (a) coherency stra<strong>in</strong> (cs): the stra<strong>in</strong> energy<br />
required to ma<strong>in</strong>ta<strong>in</strong> coherency between the (lattice<br />
mismatched) materials A and B, and (b) <strong>in</strong>terfacial energy:<br />
the energy associated with the <strong>in</strong>teractions between<br />
materials at the A=B <strong>in</strong>terface(s). To def<strong>in</strong>e these terms,<br />
it is useful to first consider the <strong>in</strong>f<strong>in</strong>ite period supercell<br />
limit N !1, for a supercell with <strong>in</strong>terface along an<br />
orientation ^G with lattice constants a k and a ? parallel<br />
and perpendicular to ^G, respectively. In this <strong>in</strong>f<strong>in</strong>iteperiod<br />
case, A=B <strong>in</strong>terfacial <strong>in</strong>teractions (which scale as<br />
the area <strong>of</strong> the <strong>in</strong>terface) contribute a negligible amount<br />
to the supercell formation energy dE sup (which scales as<br />
the volume <strong>of</strong> the superlattice):<br />
dE sup ðN !1; ^GÞ<br />
<br />
1<br />
dE CS ð^GÞ ¼m<strong>in</strong><br />
a ?<br />
2 dEepi A<br />
ða ?; a A k ; ^GÞ<br />
þ 1 <br />
2 dEepi B ða ?; a B k ; ^GÞ ;<br />
ð16Þ<br />
where dE sup is the energy <strong>of</strong> the supercell relative to<br />
equilvalent amounts <strong>of</strong> A and B <strong>in</strong> their equilibrium bulk<br />
geometries. In Eq. (16), the materials A and B are deformed<br />
<strong>in</strong> an ‘‘epitaxial’’ geometry: Both materials are<br />
brought to a common lattice constant a ? perpendicular<br />
to ^G, and the energy <strong>of</strong> each material is <strong>in</strong>dividually<br />
m<strong>in</strong>imized with respect to the lattice constant a k parallel<br />
to ^G. The epitaxial energies dE epi are the energies <strong>of</strong> A<br />
and B <strong>in</strong> these epitaxial geometries relative to their<br />
equilibrium bulk energy. For f<strong>in</strong>ite-period supercells, the<br />
energy is determ<strong>in</strong>ed not only by the coherency sta<strong>in</strong><br />
energy, but also by the <strong>in</strong>terfacial energy r times the<br />
number (2) and area (A) <strong>of</strong> these <strong>in</strong>terfaces. S<strong>in</strong>ce we are<br />
us<strong>in</strong>g energetics per atom, we must divide by the number<br />
<strong>of</strong> atoms <strong>in</strong> the cell, N. The <strong>in</strong>terfacial energy is then<br />
def<strong>in</strong>ed as<br />
dE sup ðN; ^GÞ dE sup ðN !1; bGÞ 2rð^GÞA : ð17Þ<br />
N<br />
Comb<strong>in</strong><strong>in</strong>g Eqs. (16) and (17), we see the decomposition<br />
<strong>of</strong> the supercell energy (for any period) <strong>in</strong>to stra<strong>in</strong> and<br />
<strong>in</strong>terfacial components<br />
dE SL ðN; ^GÞ ¼ 2rð^GÞA þ dE CS ð^GÞ:<br />
ð18Þ<br />
N<br />
From Eq. (18), we see that if the supercell formation<br />
energies (per atom) are plotted as a function <strong>of</strong> 1=N, the<br />
slope is just 2rA, and the y-<strong>in</strong>tercept is dE CS ð^GÞ. We<br />
have extracted the <strong>in</strong>terfacial energies from first-pr<strong>in</strong>ciples<br />
supercell calculations us<strong>in</strong>g the construction <strong>of</strong> Eq.<br />
(18). The results are shown <strong>in</strong> Fig. 6. We note that extract<strong>in</strong>g<br />
the coherency stra<strong>in</strong> energy from these calculations<br />
is not numerically stable; a more robust<br />
approach is the direct calculation <strong>of</strong> Eq. (16) described<br />
below.<br />
The tetragonal structure <strong>of</strong> h 0 precipitate embedded <strong>in</strong><br />
an fcc matrix results <strong>in</strong> partially coherent plate-shaped<br />
precipitates. The h 0 plates possess a broad coherent face